Abstract Inferences on the early evolution of the Earth's mantle can be deduced from long-lived radiogenic isotope systems such as 176Lu-176Hf and 147Sm-143Nd, for which both parent and daughter elements largely remain immobile at low metamorphic grades. However, it remains ambiguous when and to what extent mantle-crust differentiation processes had started in the Archean. For a better understanding of Archean mantle-crust evolution, we determined the initial 176Lu-176Hf, 147Sm-143Nd, and, in a new approach, the 138La-138Ce isotope compositions of a suite of Archean mafic-ultramafic rock samples from the 3.53-2.83 Ga old Pilbara Craton and 2.78-2.63 Ga old Fortescue Group in NW Australia. These rocks represent one of the best-preserved Archean successions worldwide and contain mafic-ultramafic rocks that were erupted during repeated and long-lived pulses of volcanism throughout much of the Archean. Mantle-derived mafic-ultramafic rock samples were collected from six major stratigraphic groups of the Pilbara Craton and the overlying Fortescue Group in order to characterize the parental mantle source regions of the lavas and to reconstruct the temporal evolution of the ambient mantle beneath this piece of cratonic lithosphere. In addition, we analyzed contemporaneous TTG-like igneous suites and interbedded sediments in order to reconstruct the lithospheric evolution of the Pilbara Craton. The Hf-Nd-Ce isotope data imply the onset of mantle-crust differentiation in the Pilbara Craton as early as ∼4.2 Ga, well prior to any of the preserved stratigraphy. Within error, coupled Ce-Nd-Hf isotope arrays all intersect chondritic values, implying that the Earth is of broadly chondritic composition, also for the 138La-138Ce isotope system. Mafic rocks usually yield strongly coupled eHf(i), eNd(i) and eCe(i) values that form a mixing line between an evolving depleted upper mantle composition and the primitive mantle value (eHf(i) ca. 0.0 to + 3.2, eNd(i) ca. +0.2 to +1.7 and eCe(i) ca. +0.3 to -0.1). As all Paleoarchean samples lack co-variations between Nb/Th with eHf(i) or eNd(i), contamination with an enriched crust is unlikely to explain this mixing trend. The most primitive mantle-like mafic samples show elevated GdN/YbN ratios (2.2-1.4), implying the involvement of a deep-rooted, near-primitive, upwelling mantle that was progressively mixed into the depleted upper mantle. In contrast to the mafic rocks, most, but not all komatiites are decoupled in their initial Hf-Nd-Ce isotope compositions, by having extremely radiogenic eHf(i) values at only moderately high eNd(i) and low eCe(i) values. This decoupling is best explained by the assimilation of mantle domains that underwent early melt depletion in the garnet stability field and evolved at high 176Lu/176Hf ratios but at moderate 147Sm/143Nd and 138La/138Ce ratios over time. The disappearance of rocks with decoupled Hf-Nd isotope compositions after ∼3.2 Ga is likely linked to decreasing mantle temperatures that were no longer able to melt such refractory mantle domains. Collectively, our new data for mafic rocks from the Pilbara Craton confirm the presence of long-term depleted mantle domains in the early Archean that are not sampled by the zircon Hf isotope record in the Pilbara Craton.

中文翻译：

---

皮尔巴拉克拉通玄武岩和科马提岩中 Hf-Nd-Ce 同位素系统的早期至晚期太古代地幔演化

摘要 地球地幔早期演化的推论可以从长寿命的放射性同位素系统（例如 176Lu-176Hf 和 147Sm-143Nd）中推导出来，对于这些系统，母元素和子元素在低变质等级时基本上保持不动。然而，太古宙中地幔-地壳分化过程何时以及在何种程度上开始仍然不明确。为了更好地了解太古代地幔-地壳演化，我们确定了 3.53- 3.53- 太古代地幔-超镁铁质岩石样品的初始 176Lu-176Hf、147Sm-143Nd 和 138La-138Ce 同位素组成。 2.83 Ga 老 Pilbara Craton 和 2.78-2.63 Ga 老 Fortescue Group 位于澳大利亚西北部。这些岩石代表了世界上保存最完好的太古代系列之一，其中包含在太古代大部分地区反复和长期的火山活动脉冲期间喷发的镁铁质-超镁铁质岩石。从皮尔巴拉克拉通和上覆的 Fortescue 群的六个主要地层群中收集了地幔衍生的镁铁质-超镁铁质岩石样品，以表征熔岩的母地幔源区并重建这块地幔下方环境地幔的时间演化。克拉通岩石圈。此外，我们分析了同期的类似 TTG 的火成岩套和互层沉积物，以重建皮尔巴拉克拉通的岩石圈演化。Hf-Nd-Ce 同位素数据表明皮尔巴拉克拉通的地幔-地壳分化早在~4.2 Ga 就开始了，早在任何保存的地层之前。在误差范围内，耦合的 Ce-Nd-Hf 同位素阵列都与球粒陨石值相交，这意味着地球具有广泛的球粒陨石成分，也适用于 138La-138Ce 同位素系统。基性岩通常会产生强耦合的 eHf(i)、eNd(i) 和 eCe(i) 值，它们在演化的耗尽上地幔成分和原始地幔值（eHf(i) 约 0.0 至 + 3.2， eNd(i) 大约 +0.2 至 +1.7 和 eCe(i) 大约 +0.3 至 -0.1）。由于所有古太古代样品都缺乏 Nb/Th 与 eHf(i) 或 eNd(i) 之间的协变，因此富集地壳的污染不太可能解释这种混合趋势。最原始的类似地幔的镁铁质样品显示出升高的 GdN/YbN 比率 (2.2-1.4)，这意味着涉及一种根深蒂固的、近乎原始的、上升流地幔逐渐混入耗尽的上地幔中。与镁铁质岩石相反，大多数（但并非所有）科马提岩在其初始 Hf-Nd-Ce 同位素组成中是解耦的，因为它们具有极高的放射成因 eHf(i) 值，且只有中等高的 eNd(i) 和较低的 eCe(i) 值. 这种解耦最好通过地幔域的同化来解释，这些地幔域在石榴石稳定场中经历了早期熔体消耗，并随着时间的推移以高 176Lu/176Hf 比率但以中等 147Sm/143Nd 和 138La/138Ce 比率演化。大约 3.2 Ga 之后具有解耦 Hf-Nd 同位素组成的岩石的消失可能与地幔温度降低有关，而地幔温度不再能够熔化这种难熔的地幔域。总的来说，